Splitting the random wierd stuff off into a seperate distribution wouldn't do much to improve the speed of releases. If you look at what's holding us up you'll see that it's key things like not having any boot floppies.
If there is a critical problem with something obscure then it won't hold the release up if it doesn't get fixed - the package will simply get pulled. For things like boot floppies that we really need that is just not possible. Removing the small packages wouldn't get anything done faster - if people wished to spend their time on the things that are holding us up, they would be doing so already.
We do need to rethink the way we do released (or rather, find people willing to work on the important bits that nobody seems to have worked on) but pulling packages out of the distribution is not the answer. Take a look through the mailing list archives to see the ideas that have been floated - the main one appears to be to have a semi-stable distribution which contains packages that have been tested in unstable and which are officially BugFree(tm).
One thing to do would be to look for addresses inside the tarballs. Your statistics reflect who is in charge of projects rather than who is contributing to them. It is unsurprising that people only tend to be the primary author of a few programs - it would be much more interesting to see how many people were sending patches and so on.
Of course, it's much harder to collect those statistics.
Creative naming schemes are fun when your environment is small. They don't scale though.
For situations like this you tend to have lots of machines that are pretty much identical aside from the hostname. These machines can usefully be named "foo0000" or whatever.
The more distinct names tend to be useful mainly when the machines are fairly unique and can be easily distinguished.
Indeed - reading the bit on the front page I had thought that that was going to be complaining about it being a.gov domain, not about where it is in.gov. There is a difference between a political party and the apparatus of government and I'm surprised that the people handing out the address didn't notice or think of that (it's one of the first things in the RFC).
If it does have to be under.gov, then gop.gov doesn't seem unreasonable - I would imagine that there are Republicans in places other than Congress which tends to suggest that putting them under house.gov is a bad idea. If they have to be under.gov, gop.gov seems the lesser of two evils. I could be missing something about American politics, though.
More than that, much of the target audience is not really competant to asses the arguments. To a huge proportion of people out there the entire discussion is basically meaningless - they don't understand any of the technical terms being thrown about.
If there are people who happen to believe what's said or don't see the other side mentioned at all then that's a plus for them, but the main thing is creating doubt in the minds of the general audience. To these people the very existance of a discussion is cause for concern.
What do you mean by Beowulf support? Software for clusters is almost always custom-written - there's very little it would do for most "regular" applications without work. To a certain extent, any distribution supporting networking provides basic tools for a Beowulf.
What they could do would be to ship packages of common building block libraries like PVM or an MPI implementation and provide tools for managing large clusters of machines, but none of that would do anything for most people who didn't want to write custom apps (pvmpov is the only relativley general pre-written one I can think of off-hand).
Indeed. Most of the (public - I've no idea what turns up in his inbox, but I imagine it's worse) criticism of ESR is of the form "how dare you say that I believe that". Like someone said earlier on talking about "Church and state", Eric often expresses his views as being the views of everyone and sometimes seems to be advocating as much to other people in the community as to people outside it. This rubs those who disagree with him up the wrong way, and with reason.
On the other hand, he can't really go around adding riders to everything he says and covering all viewpoints - it would create confusion and make him completely ineffective. A few mentions or weasel words and more consultation would go some way, but ultimately it's unavoidable and people are going to have to accept that if there is someone like Eric, they may end up not telling the whole story the whole time. Everyone (not just Eric) needs to learn that there are other viewpoints and that there are some issues which come down to individual choice.
One thing that would help a lot would be a less abrasive response from Eric to disagreements. Often, what happens (again, in public) is that any disagreement is met with a "you don't understand, you're wrong, I'm right, I'm in charge, get out of my way" style response. Perhaps understandable given the attitude he sometimes seems to get, but hardly likely to solve anything. Yes, other people do it to - but generally Eric is the one with the highest profile.
If your lecturers are any good then you shouldn't want to zone out in class. Even if they aren't, pretty much every lecturer I've had (including those who give out full printed notes in advance) talks around the notes rather than simply reading them out.
> Sorry to disappoint you, but Linux runs on just about everything NetBSD does plus Palm Pilots! And there were less-publicized ones for *BSD which means many users probably still don't know. Atleast the Linux developers can admit to
This is unfair. I see the security announcements for NetBSD and OpenBSD, and they seem to be reasonably prompt and full. It's another part of the "Linux gets more publicity" thing - there's not really much they can do beyond their own security announce list and places like bugtraq.
The point with security is not that *BSD have no security flaws (although I'd imagine OpenBSD is getting pretty near that by now:-) ) it's that because there are more Linux systems out there crackers are more likely to try to exploit problems in Linux. A form of security through obscurity, perhaps - but it's a sensible enough point.
Definately. Unix isn't perfect by any strech of the imagination. It may be one of the better systems out there, but for much the same reason that democracy is - everything else sucks more.
Acknowledging the existance of problems is not the same as ignoring them. Without looking at what's wrong it's hard to improve.
> Perhaps RedHat could have emphasized this more, or pointed out that it's an SEC regulation.
Definately - it's the same problem as with the US residency reqirement. It read like an advert with small print rather than something they were trying to make useful.
Russ Allbery's A Rant. I don't know that it's profoundly affected anything, but there's a good chunk of what's good and bad about Usenet and why anyone should care in there.
> First, it's not spam--If you were sent a username and password for an online trading account worth a million bucks unsolicited in the e-mail (with a reason and your name attached) would you turn it down?
It's just as much spam as the other junk I get - it's unsolicited, it's commercial and it's e-mail.
The fact that it contains some account information is irrelevant - porn sites sometimes do this, and they at least would allow me to use their services (the offer is open only to US residents, which I am not).
In any case, what standing in line? In all the UK elections I've voted in there's been no problem just walking into the polling stating - not a queue in sight.
I understand that the US has generally more complex ballot papers, but even so...
In British elections it is already possible to discover who voted how. There is a number on the ballot paper which gets recorded on the electoral register when you vote.
Of course, it would be very time consuming to go back and work out the matches - but it is possible.
>.. has a similiar directory structure for their RPMS. Why does this make Debian superior?
Most people are somewhat i386 centric? Debian is rather anal about licenses.
> 1. how does one upgrade a debian box if a security issue is found with a package?
Either grab the deb and install it manually, or use dselect/apt to work out if there are any updates and install them (see next question).
> 2. What exactly does apt-get update actually do (it seems to just change a few gzipped files on my machine representing the directory structure of the debian ftp site)
That's what it does - it updates the list of avalible packages. To upgrade all the upgradeable packages, use "apt-get upgrade". To upgrade/install a specfic package, use "apt-get install foo".
> Does ir get packages which have been changed due to security related issues?
No. If you have the security updates archive locations (security.debian.org and proposed-updates) in your sources file, then doing "apt-get update ; apt-get upgrade" should do the right thing.
> 3. DOes anyone have a script/howto for making debs. RPMS seem really easy, but the stuff for debs on the debian site seemed a litte too confusing.
I found the easiest thing was to just do it. It's sometimes instructive to look at the diff files of existing packages to see how they do things. A good starting point is to use dh_make to put in a skeleton which works for packages configured using autoconf (ISTR the potato version broke - if it's still broken, try the one from slink). Use lintian to check for errors in the built debs - it's very useful.
> 4. Has the debian open-source manual been released, yet?
Pass. There's a whole bunch of debian manuals - take a look at the Debian Documentation Project for more info (it's linked from the devel section of the web site).
I'll second the art-hose recommendation - there's an awful lot of excellent films being made, but I can't really think of any that hit the multiplexes round here.
What I would add is that Lucas has been doing the cash in thing for quite some time. Just prior to the video release of the remastered Star Wars, he pulled all prints of the films, preventing theatrical screenings.
"rich latencies"?!?!, "Communitarian Software"
on
The Power of Openness
·
· Score: 1
There's a few bits of that and a few bits of badly used terminology, but don't knock him too hard. Mostly it's perfectly readable, and it's certainly a model for writing from academic people from the arts.
> He seemed to be right on the money about Stallman's beliefs, although he obviously didn't agree with them.
It's all there, but deemphasiasised in such a way that it's a bit misleading. The ability to take free software and make a closed sourced product with it is aluded to, but never explicitly stated. This makes the reasoning behind RMS' position much more confused than it need be.
As you say the author is biased, and as a result the article does a pretty good job of rubbishing RMS.
*Really* nasty reviews can be very entertaining, though. Things like the Bookwork column in the UK magazine Private Eye are fantastic. It's not just a case of bad notices - it's really brutal but very accurate (well written and argued) reviewing.
I take the point about making money from sales, but perhaps if anyone is writing really acid reviews you could include them for sheer entertainment value.
Good job I tried to view this using Lynx the first time round - the document selects Tacoma and Ariel for fonts, and my Netscape doesn't deal at all well with their absence, rendering them in microscopic and ugly courier.
I can't install xfstt on all the machines I use - some don't have Windows licenses and some I don't have any control over.
> I guess I am alone in thinking this movie is an over-hyper, under-written, and poorly plotted movie.
The plot and content aren't the point. They really don't stand up to much examination by themselves. What is remarkable is the direction and design of the movie - they are very much unusual. If you let yourself be drawn into the film they are very effectlve.
To me, it's more about taking you into a state of mind than anything else.
MS have a monopoly in a certain sector of the desktop OS market. To do an awful lot of things you have no option but to run Windows, either because of software avalibility or because explaining the difference to new users is just too much effort. While the various Unix variants are very nice, for a random person buying a PC for home use the market is such that they don't really make sense.
MS isn't really doing anything visibly exciting with core products like 9[58] or Office - I haven't noticed anything much new that I wanted to use in Word or Excel since 6.0. They still release new versions every once in a while, but that's not the same thing. Merely providing more features doesn't sound like innovation to me - it's still a Widget, even if it is Mk.4 Widget with knobs on.
With regard to Apple, they had some big management problems. They moved to PowerPC and IIRC did some stuff with the OS, but didn't manage to bring the users along with them (high initial costs and all that).
Slashdot Mirror
Splitting the random wierd stuff off into a seperate distribution wouldn't do much to improve the speed of releases. If you look at what's holding us up you'll see that it's key things like not having any boot floppies.
If there is a critical problem with something obscure then it won't hold the release up if it doesn't get fixed - the package will simply get pulled. For things like boot floppies that we really need that is just not possible. Removing the small packages wouldn't get anything done faster - if people wished to spend their time on the things that are holding us up, they would be doing so already.
We do need to rethink the way we do released (or rather, find people willing to work on the important bits that nobody seems to have worked on) but pulling packages out of the distribution is not the answer. Take a look through the mailing list archives to see the ideas that have been floated - the main one appears to be to have a semi-stable distribution which contains packages that have been tested in unstable and which are officially BugFree(tm).
One thing to do would be to look for addresses inside the tarballs. Your statistics reflect who is in charge of projects rather than who is contributing to them. It is unsurprising that people only tend to be the primary author of a few programs - it would be much more interesting to see how many people were sending patches and so on.
Of course, it's much harder to collect those statistics.
Creative naming schemes are fun when your environment is small. They don't scale though.
For situations like this you tend to have lots of machines that are pretty much identical aside from the hostname. These machines can usefully be named "foo0000" or whatever.
The more distinct names tend to be useful mainly when the machines are fairly unique and can be easily distinguished.
Indeed - reading the bit on the front page I had thought that that was going to be complaining about it being a .gov domain, not about where it is in .gov. There is a difference between a political party and the apparatus of government and I'm surprised that the people handing out the address didn't notice or think of that (it's one of the first things in the RFC).
.gov, then gop.gov doesn't seem unreasonable - I would imagine that there are Republicans in places other than Congress which tends to suggest that putting them under house.gov is a bad idea. If they have to be under .gov, gop.gov seems the lesser of two evils. I could be missing something about American politics, though.
If it does have to be under
More than that, much of the target audience is not really competant to asses the arguments. To a huge proportion of people out there the entire discussion is basically meaningless - they don't understand any of the technical terms being thrown about.
If there are people who happen to believe what's said or don't see the other side mentioned at all then that's a plus for them, but the main thing is creating doubt in the minds of the general audience. To these people the very existance of a discussion is cause for concern.
What do you mean by Beowulf support? Software for clusters is almost always custom-written - there's very little it would do for most "regular" applications without work. To a certain extent, any distribution supporting networking provides basic tools for a Beowulf.
What they could do would be to ship packages of common building block libraries like PVM or an MPI implementation and provide tools for managing large clusters of machines, but none of that would do anything for most people who didn't want to write custom apps (pvmpov is the only relativley general pre-written one I can think of off-hand).
Indeed. Most of the (public - I've no idea what turns up in his inbox, but I imagine it's worse) criticism of ESR is of the form "how dare you say that I believe that". Like someone said earlier on talking about "Church and state", Eric often expresses his views as being the views of everyone and sometimes seems to be advocating as much to other people in the community as to people outside it. This rubs those who disagree with him up the wrong way, and with reason.
On the other hand, he can't really go around adding riders to everything he says and covering all viewpoints - it would create confusion and make him completely ineffective. A few mentions or weasel words and more consultation would go some way, but ultimately it's unavoidable and people are going to have to accept that if there is someone like Eric, they may end up not telling the whole story the whole time. Everyone (not just Eric) needs to learn that there are other viewpoints and that there are some issues which come down to individual choice.
One thing that would help a lot would be a less abrasive response from Eric to disagreements. Often, what happens (again, in public) is that any disagreement is met with a "you don't understand, you're wrong, I'm right, I'm in charge, get out of my way" style response. Perhaps understandable given the attitude he sometimes seems to get, but hardly likely to solve anything. Yes, other people do it to - but generally Eric is the one with the highest profile.
If your lecturers are any good then you shouldn't want to zone out in class. Even if they aren't, pretty much every lecturer I've had (including those who give out full printed notes in advance) talks around the notes rather than simply reading them out.
Doing this can make your life very much easier.
> Sorry to disappoint you, but Linux runs on just about everything NetBSD does plus Palm Pilots! And there were less-publicized ones for *BSD which means many users probably still don't know. Atleast the Linux developers can admit to
This is unfair. I see the security announcements for NetBSD and OpenBSD, and they seem to be reasonably prompt and full. It's another part of the "Linux gets more publicity" thing - there's not really much they can do beyond their own security announce list and places like bugtraq.
The point with security is not that *BSD have no security flaws (although I'd imagine OpenBSD is getting pretty near that by now :-) ) it's that because there are more Linux systems out there crackers are more likely to try to exploit problems in Linux. A form of security through obscurity, perhaps - but it's a sensible enough point.
Definately. Unix isn't perfect by any strech of the imagination. It may be one of the better systems out there, but for much the same reason that democracy is - everything else sucks more.
Acknowledging the existance of problems is not the same as ignoring them. Without looking at what's wrong it's hard to improve.
> Perhaps RedHat could have emphasized this more, or pointed out that it's an SEC regulation.
Definately - it's the same problem as with the US residency reqirement. It read like an advert with small print rather than something they were trying to make useful.
Russ Allbery's A Rant. I don't know that it's profoundly affected anything, but there's a good chunk of what's good and bad about Usenet and why anyone should care in there.
> First, it's not spam--If you were sent a username and password for an online trading account worth a million bucks unsolicited in the e-mail (with a reason and your name attached) would you turn it down?
It's just as much spam as the other junk I get - it's unsolicited, it's commercial and it's e-mail.
The fact that it contains some account information is irrelevant - porn sites sometimes do this, and they at least would allow me to use their services (the offer is open only to US residents, which I am not).
In any case, what standing in line? In all the UK elections I've voted in there's been no problem just walking into the polling stating - not a queue in sight.
I understand that the US has generally more complex ballot papers, but even so...
In British elections it is already possible to discover who voted how. There is a number on the ballot paper which gets recorded on the electoral register when you vote.
Of course, it would be very time consuming to go back and work out the matches - but it is possible.
> .. has a similiar directory structure for their RPMS. Why does this make Debian superior?
Most people are somewhat i386 centric? Debian is rather anal about licenses.
> 1. how does one upgrade a debian box if a security issue is found with a package?
Either grab the deb and install it manually, or use dselect/apt to work out if there are any updates and install them (see next question).
> 2. What exactly does apt-get update actually do (it seems to just change a few gzipped files on my machine representing the directory structure of the debian ftp site)
That's what it does - it updates the list of avalible packages. To upgrade all the upgradeable packages, use "apt-get upgrade". To upgrade/install a specfic package, use "apt-get install foo".
> Does ir get packages which have been changed due to security related issues?
No. If you have the security updates archive locations (security.debian.org and proposed-updates) in your sources file, then doing "apt-get update ; apt-get upgrade" should do the right thing.
> 3. DOes anyone have a script/howto for making debs. RPMS seem really easy, but the stuff for debs on the debian site seemed a litte too confusing.
I found the easiest thing was to just do it. It's sometimes instructive to look at the diff files of existing packages to see how they do things. A good starting point is to use dh_make to put in a skeleton which works for packages configured using autoconf (ISTR the potato version broke - if it's still broken, try the one from slink). Use lintian to check for errors in the built debs - it's very useful.
> 4. Has the debian open-source manual been released, yet?
Pass. There's a whole bunch of debian manuals - take a look at the Debian Documentation Project for more info (it's linked from the devel section of the web site).
I'll second the art-hose recommendation - there's
an awful lot of excellent films being made, but
I can't really think of any that hit the multiplexes
round here.
What I would add is that Lucas has been doing
the cash in thing for quite some time. Just
prior to the video release of the remastered
Star Wars, he pulled all prints of the films,
preventing theatrical screenings.
There's a few bits of that and a few bits of badly used terminology, but don't knock him too hard. Mostly it's perfectly readable, and it's certainly a model for writing from academic people from the arts.
No, he says they may be the only company selling CDs containing only free software. Debian don't sell CDs.
He's probably wrong on that too, but anyway.
> He seemed to be right on the money about Stallman's beliefs, although he obviously didn't agree with them.
It's all there, but deemphasiasised in such a way that it's a bit misleading. The ability to take free software and make a closed sourced product with it is aluded to, but never explicitly stated. This makes the reasoning behind RMS' position much more confused than it need be.
As you say the author is biased, and as a result the article does a pretty good job of rubbishing RMS.
*Really* nasty reviews can be very entertaining, though. Things like the Bookwork column in the UK magazine Private Eye are fantastic. It's not just a case of bad notices - it's really brutal but very accurate (well written and argued) reviewing.
I take the point about making money from sales, but perhaps if anyone is writing really acid reviews you could include them for sheer entertainment value.
Good job I tried to view this using Lynx the first time round - the document selects Tacoma and Ariel for fonts, and my Netscape doesn't deal at all well with their absence, rendering them in microscopic and ugly courier.
I can't install xfstt on all the machines I use - some don't have Windows licenses and some I don't have any control over.
> I guess I am alone in thinking this movie is an over-hyper, under-written, and poorly plotted movie.
The plot and content aren't the point. They really don't stand up to much examination by themselves. What is remarkable is the direction and design of the movie - they are very much unusual. If you let yourself be drawn into the film they are very effectlve.
To me, it's more about taking you into a state of mind than anything else.
Flame bait, I know.... but anyway:
MS have a monopoly in a certain sector of the desktop OS market. To do an awful lot of things you have no option but to run Windows, either because of software avalibility or because explaining the difference to new users is just too much effort. While the various Unix variants are very nice, for a random person buying a PC for home use the market is such that they don't really make sense.
MS isn't really doing anything visibly exciting with core products like 9[58] or Office - I haven't noticed anything much new that I wanted to use in Word or Excel since 6.0. They still release new versions every once in a while, but that's not the same thing. Merely providing more features doesn't sound like innovation to me - it's still a Widget, even if it is Mk.4 Widget with knobs on.
With regard to Apple, they had some big management problems. They moved to PowerPC and IIRC did some stuff with the OS, but didn't manage to bring the users along with them (high initial costs and all that).